The EMBO Journal vol.14 no.19 pp.4860-4871, 1995

A common maturation pathway for small nucleolar

Michael P.Terns1 2, Christian Grimm, mRNA (Birmstiel and Schaufele, 1988). Most nucleo- Elsebet Lund and James E.Dahlberg plasmic snRNAs are made by RNA polymerase II (RNAP II) and contain a sequence element known as the Sm site Department of Biomolecular Chemistry, 1300 University Avenue, to which the group of Sm proteins bind. After binding of University of Wisconsin, Madison, WI 53706, USA the Sm proteins to these RNAs their 5' m7G caps undergo 'Present address: Department of and Molecular , hypermethylation to trimethylguanosine 5' cap structures. Life Sciences Building, University of Georgia, Athens, GA 30602- The spliceosomal U6 RNA, which is made by RNAP III 7229, USA does not have an m7G cap nor an Sm protein binding site. 2Corresponding author snoRNAs are synthesized by RNAP II (i.e. U3, U8 and U13) or RNAP III (i.e. 7-2/MRP and plant U3) and some We have shown that precursors of U3, U8 and U14 are processed from the intronic sequences of mRNAs (i.e. small nucleolar RNAs (snoRNAs) are not exported to U14-U22). Specific steps of pre-rRNA processing require the cytoplasm after injection into Xenopus oocyte nuclei particular snoRNAs such as U3, U8, U14, 7-2/MRP and but are selectively retained and matured in the nucleus, U22 RNAs (Tyc and Steitz, 1989; Kass et al., 1990; Li where they function in pre-rRNA processing. Our et al., 1990; Savino and Gerbi, 1990; Hughes and Ares, results demonstrate that Box D, a conserved sequence 1991; Peculis and Steitz, 1993, 1994; Morrissey and element found in these and most other snoRNAs, Tollervey, 1995). Most nucleolar RNAs contain common plays a key role in their nuclear retention, 5' cap sequence elements (Boxes C', C and D) and are associated hypermethylation and stability. Retention of U3 and with the conserved 34-36 kDa nucleolar protein fibrillarin U8 RNAs in the nucleus is saturable and relies on one (Caizergues-Ferrer et al., 1991; Tollervey et al., 1991). or more common factors. Hypermethylation of the 5' Mature snoRNAs contain either a trimethylguanosine 5' caps of U3 RNA occurs efficiently in oocyte nuclear cap structure (Tyc and Steitz, 1989) or a triphosphate 5' extracts lacking nucleoli, suggesting that precursor end (Kiss et al., 1991; Yuan and Reddy, 1991) depending snoRNAs are matured in the nucleoplasm before they on whether they are RNAP II or III gene products, are localized to the nucleolus. Surprisingly, m7G- respectively, but those that are processed from mRNA capped precursors of spliceosomal small nuclear RNAs introns contain monophosphate 5' ends (Kiss and (snRNAs) such as pre-Ul and U2, can be hypermethyl- Filipowicz, 1993; Tycowski et al., 1993). ated in nuclei if the RNAs are complexed with Sm snRNAs must undergo maturation in order to become proteins. This raises the possibility that a single nuclear functional. The nucleoplasmic snRNAs bind the common hypermethylase activity may act on both nucleolar and Sm proteins and undergo hypermethylation of their m7G spliceosomal snRNPs. cap structures after being exported to the cytoplasm Keywords: nucleolar snoRNA/nucleus/snRNA/transport/ (Mattaj, 1986). Recent evidence indicates that export of Xenopus oocyte nucleoplasmic pre-snRNAs to the cytoplasm is facilitated by interactions of specific nuclear factors with the m7G cap and sequence domains within the body of these RNAs (Izaurralde et al., 1992; Terns et al., 1993a,b; Jarmolowski Introduction et al., 1994). Likewise, import of mature snRNAs back Small nuclear RNAs (snRNAs), together with their into the nucleus is promoted by the bound Sm proteins associated proteins, are required to process precursors of and the trimethylguanosine 5' cap structures (Fischer and other cellular RNAs into mature RNA species (for review, Luhrmann, 1990; Hamm et al., 1990; Fischer et al., 1993). see Baserga and Steitz, 1993). The many different snRNAs The maturation pathways used by snoRNAs have not can be generally divided into two classes based on the been extensively studied. Despite the similarities between intranuclear location of their function. Nucleoplasmic U3 RNA and nucleoplasmic snRNAs (both are synthesized snRNAs function in pre-mRNA processing in the nucleo- in the nucleus by RNAP II with an m7G cap that later plasm (reviewed in Luhrmann et al., 1990; Green, 1991), gets hypermethylated), the maturation pathways used by whereas nucleolar snRNAs (snoRNAs) function in pre- U3 RNA and nucleoplasmic snRNAs differ significantly. rRNA processing (or ribosome biogenesis) in the nucleolus Recently, we showed that U3 RNA is not exported to the (reviewed in Filipowicz and Kiss, 1993; Fournier and cytoplasm but acquires its trimethylguanosine cap structure Maxwell, 1993; Maxwell and Foumier, 1995) in the nucleus (Tems and Dahlberg, 1994). The nucleoplasmic snRNAs include the spliceosomal In this study we show that in addition to U3 snRNA, snRNAs (i.e. Ul, U2, U4, U5 and U6 RNAs) which are other snoRNAs including U8 and U14 RNAs are matured required for intron removal from pre-mRNAs (Guthrie solely within the nucleus. We also demonstrate that Box and Patterson, 1988; Luhrmann et al., 1990) and U7 RNA D, a six nucleotide sequence element present near the 3' which is necessary for 3' end formation of histone pre- terminus of most snoRNAs (Filipowicz and Kiss, 1993;

486040 Oxford University Press Maturation of small nucleolar RNAs

A T- tz > B h :---m immunoprecipitation of nuclear RNAs with antibodies N ,\ (- C- (- 1) s p s specific to the monomethyl (m7G) and trimethyl (m2'2'7G) guanosine cap structures (present on precursor and mature t-3_I snRNAs, respectively). Thus, m7G-capped precursors of tU3 ___l_ * snoRNAs such as U3 and U8 are not exported to the _2ot....ei cytoplasm but are matured solely within the nucleus. UJ2l.__. .mw1 Moreover, these RNAs were both associated with fibrillarin protein following their injection into nuclei (data not w. shown; Peculis and Steitz, 1994). ___ L'I w__ C".l_R1 ..... Structural elements needed for nuclear retention _ of snoRNAs _ To determine if the nuclear retention of U3 and U8 RNAs U8 _ _ U8X_...... 8. was saturable, we assayed the nucleocytoplasmic distribu- tion of the RNAs when they were present in nuclei in high amounts. Injection of 0.1 ng of U3 DNA template resulted in accumulation ofU3 RNA only in the nucleus (Figure 2A, Fig. 1. Nuclear retention and hypermethylation of snoRNAs. lanes 1 and 2). However, injection of 10 times that amount (A) Nucleocytoplasmic distribution of U3, U2, Ul and U8 RNAs following their injection into oocyte nuclei. 32P-labeled m7G-capped of DNA, to elevate the level of U3 RNA produced, resulted U3, U2, Ul and U8 RNAs were synthesized in vitro and the RNA in the accumulation of significant amounts of U3 RNA in mixture was injected into nuclei of Xenopus oocytes. After 2 and 6 h the cytoplasm (lanes 3 and 4). Moreover, the percentage of of incubation at 18°C, the labeled RNAs present in the nuclear (N) U3 RNA that was trimethylguanosine-capped decreased and cytoplasmic (C) fractions of the oocytes were isolated and from >90% (Tems and Dahlberg, 1994) to -50% (Figure analyzed by electrophoresis in a denaturing polyacrylamide gel. Lane 1 (M) shows the RNAs prior to injection. (B) Identification of the 5' 2B, lanes 1-4) when high amounts of U3 DNA were cap structure (either m7G or m2.2'7G) by immunoprecipitation. injected. These results indicate that oocytes have a limited Precipitation was carried out with nuclear RNAs shown in lane 4 of capacity for both retaining and hypermethylating U3 (A), using anti-m7G (Munns et al., 1982) or anti-m 2'27G (Bringmann snoRNA in the nucleus. et al., 1983) cap antibodies as indicated. The RNAs present in the It is that the appearance of U3 RNA in the total sample (T), precipitate (P) and supematant (S) fractions were unlikely separated by gel electrophoresis as in (A). cytoplasm was the result of non-specific RNA leakage during oocyte fractionation since co-expressed U6 RNA (an RNA which also does not exit the nucleus) remained Foumier and Maxwell, 1993; Maxwell and Fournier, exclusively within the nucleus (Figure 2A, lanes 3 and 1995), is necessary for the efficient nuclear hypermethyl- 4). Furthermore, as shown in lanes 4 and 6, the appearance ation of these RNAs both in vivo and in vitro. The retention of U3 RNA in the cytoplasm was prevented by injection of of U3 and U8 RNAs in the nucleus is a saturable process an antibody that blocks RNA export (Terns and Dahlberg, and a common titratable factor is apparently involved in 1994; E.Lund, unpublished data) by binding to nuclear the nuclear retention of both RNAs. Box D sequences are pore glycoproteins. Essentially all of the U3 RNA that essential for the retention of U8 RNA but not U3 RNA. appeared in the cytoplasm under conditions of elevated We conclude that specific interactions between precursors U3 RNA synthesis was m7G-capped (Figure 2B, compare of snoRNAs and limiting components in the nucleus are U3 and Ul RNAs in lanes 5-8). Retention of U8 RNA responsible for retaining these RNAs in the nucleus. was also a saturable process as shown by its appearance in the cytoplasm when large amounts of the RNA were injected into nuclei (see below). Results Since both U3 and U8 RNAs are retained within the Nuclear retention and hypermethylation of capped nucleus, we tested whether they shared common saturable snoRNAs retention mechanisms. To do this we co-injected radio- To test if nuclear retention and cap hypermethylation are labeled U3 and U8 RNAs with increasing amounts of general properties of snoRNAs made by RNAP II rather these RNAs as non-radioactive competitors and monitored than specific behaviors of U3 RNA (Tems and Dahlberg, the nucleocytoplasmic distributions of radiolabeled RNAs 1994), we monitored the intracellular localization of (Figure 3A). As a control for the capacity of oocyte nuclei another snoRNA, U8 RNA (Figure lA). The nucleocyto- to retain RNA, we co-injected U6 RNA. plasmic distributions of Xenopus U8 RNA and control In the absence of competitor RNA, all of the labeled spliceosomal snRNAs that had been co-injected into the U3 and U8 RNA remained in the nucleus as expected nuclei of Xenopus oocytes, were determined by polyacryl- (lanes 2 and 3). When increasing amounts of unlabeled amide gel electrophoresis (PAGE) of the RNAs isolated competitor U3 RNA were co-injected, both U3 and U8 from the nuclear (lanes 2 and 4) and cytoplasmic (lanes RNAs accumulated in the cytoplasm in a dose-dependent 3 and 5) fractions. Both U8 and U3 RNA remained in the (compare lanes 5, 7 and 9) and time-dependent (compare nucleus whereas spliceosomal snRNAs U2 and Ul were lanes 8-9 with 10-11) manner. U8 RNA was also an exported to the cytoplasm (compare RNAs in lanes 4 effective competitor of its own nuclear retention but it did and 5). not have a great effect on retention of U3 RNA lanes 12- Despite not being exported from the nucleus, the m7G- 15). In the presence of high levels of U3 or U8 competitor capped precursor of U8 RNA became efficiently hyper- RNA, both of these RNAs were less stable and underwent methylated (Figure iB, lanes 4 and 5) as assayed by trimming at their 3' ends.

4861 M.P.Terns et aL

nn dti .41-, *\. ...'LI.1.asI k R

_,,7t...... - ., _I

1 1 ... Ili M7: 44... .. iw I'O

Fig. 2. Saturation of nuclear retention of U3 RNA. (A) Effect of increased U3 gene expression on nuclear retention of U3 RNA. U3 genes [0.1 ng (1 X) or 1.0 ng (lOX) of DNA] were co-injected with Ul and U6 genes (1.0 and 0.1 ng of DNA, respectively) into nuclei of oocytes in the absence (-) and presence (+) of 150 ng of an anti-nuclear pore complex antibody (mAb 414) (Davis and Blobel, 1986). After h of incubation, [a32P]GTP was injected into cytoplasms and the newly-made snRNAs were labeled for 6 h. 32P-labeled RNAs present in the nuclear (N) and cytoplasmic (C) fractions were analyzed by gel electrophoresis. The autoradiographic exposure of U3 and U6 RNAs in lanes 1 and 2 was 12 times longer than that of lanes 3-6. Also, since high levels of U3 genes competed for transcription of co-injected Ul genes, the exposure for Ul RNA in lanes 3-6 was -1.5 times longer than that of lanes 1 and 2. The positions of U3, U6 and the precursor (m7G-capped) and mature (m2'2'7G-capped) forms of U1 RNA are indicated. (B) 5' cap analysis of nuclear and cytoplasmic snRNAs made in oocytes injected with high levels (lOX) of U3 genes. The Ul and U3 RNA present in the nuclear (A; lane 3) and cytoplasmic (A; lane 4) fractions were precipitated with anti-m7G or anti-m2'2'7G antibodies as indicated. RNAs were purified from both the precipitate (P) and supematant (S) fractions and analyzed by gel electrophoresis.

4 t

!.'|' i;.

_

\ .;

': _:

I-XI...-... *.... *. - l.I_ ..- - _.t *. .------._. t6!- ^ ..^...... ^..w...

i _ L6._ _ ..PI }i'R.1 .. w, w:

Fig. 3. U3 and U8 RNAs compete for retention sites within the nucleus. (A) Effect of competitor U3 and U8 RNAs on the nucleocytoplasmic distribution of snRNAs. 32P-labeled in vitro-made, m7G-capped U3, U8 and y-mpppG-capped U6 RNAs were injected into nuclei of oocytes in the absence (lanes 2 and 3) or presence (lanes 4-15) of unlabeled competitor m7G-capped U3 (lanes 4-11) or U8 RNAs (lanes 12-15). The levels of injected competitor RNAs (in fmol) were 25 (lanes 4 and 5), 75 (lanes 6 and 7) and 225 (lanes 8-15). After 2.5 or 8 h, the labeled RNAs present in the nuclear (N) and cytoplasmic (C) oocyte fractions were isolated and analyzed by gel electrophoresis. U6 RNA, which normally does not exit the nucleus, served as the control for imperfect oocyte fractionation and/or partial injection of the RNAs into the cytoplasm; the low levels of U6 RNA observed in the cytoplasmic fractions of lanes 9, 13 and 15 were not observed consistently in all experiments.ILane 1 (M)_shows the RNAs prior to injection. (B) Effect of competitor Ul RNA on the nucleocytoplasmic distribution of snRNAs. Injections were carried out as described in (A) except that the labeled RNAs were co-injected in the absence (lanes 1 and 2) or presence (lanes 3 and 4) of 225 fmol of unlabeled m7G-capped U1 competitor RNA and labeled U1 RNA was also analyzed. Lane S (M) shows the RNAs prior to injection.

The specificity of snoRNA retention is shown by the shortening of U 1 3' ends (Neuman de Vegvar and lack of effect of the competitor RNAs on the nuclear Dahlberg, 1990). retention of U6 RNA (lanes 10-11 and 14-15) or the rate We asked if the Box D sequence element, which is of U2 RNA export (data not shown). Also, co-injection present in nearly all snoRNAs (Figure 4A), is important of a comparable amount of unlabeled Ul RNA did not for the nuclear retention and stability of U3 and U8 RNAs. affect the retention of either U3 or U8 in the nucleus Surprisingly, U3 and U8 RNAs with Box D substitutions (Figure 3B, compare lanes 1 and 3). As shown in lanes 2 behaved differently from each other. Mutant U3 RNA and 4, the presence of competitor U1 RNA resulted in 3' was retained in the nucleus, whereas mutant U8 RNA trimming of Ul RNA and significantly retarded the import accumulated in the cytoplasm (Figure 4B, lanes 3 and 5). but not export of Ul RNA, presumably due to the Thus, the Box D element is necessary for the retention of

4862 Maturation of small nucleolar RNAs

A t-^rt

A;{.fOC

E!-C, G C, t, VIUl:~~~~~~~~~~~~~~~~~~~~~~r-Ct*

{i

A-1i X :1 ^r t; C i

x c C., c 1 -i w .,

AG A U c c < ^ '.^ ~ DtBo ';r)ac ,j''I Ste

A. o D Ti q -0A. U1 0-t.-,,, 3' Stem

220 U3 U8

B C U3 2.0hr 6.0Ohr 1 2.0hr 6.0hr 1 M I Cl N Cl M I N N r~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 5(Ca ClF1 C1 U3M 4w mZ27GpppG U2 ApppG !mto U1 x pppG -Il- w U8AD MWW OR m7GpppG

2 3 4 5 1 2 3 4 5

Fig. 4. Effect of Box D sequences and the trimethylguanosine 5' cap on nuclear retention of snoRNAs. (A) Proposed secondary structures of U3, U8 and U14 snoRNAs. Secondary structure models of X.laevis U3A (Savino et al., 1992), X.laevis U8 (Peculis and Steitz, 1993) and mouse U14.5 (Shanab and Maxwell, 1991) RNAs are shown. Sequences of Box D (U3, U8 and U14) and 3' terminal stem structures (U3 and U14) are boxed. 5' terminal nucleotides present in in vitro synthesized RNAs but not in endogenous RNAs are in parentheses. (B) Nucleocytoplasmic distributions of mutant U3 and U8 RNAs that lack Box D sequences. U3 and U8 variants that have sequences substituted for Box D were synthesized in vitro with m7G caps and co-injected into nuclei of oocytes. After incubation for the times indicated, the distributions of the RNAs in nuclear (N) and cytoplasmic (C) fractions were analyzed by gel electrophoresis. The m7G-capped precursors of Ul and U2 RNAs were co-injected as controls for RNAs that are normally exported to the cytoplasm. Lane 1 (M) shows the RNAs prior to injection. (C) Nuclear retention of U3 RNAs containing different 5' cap structures. 32P-labeled wild-type U3 RNAs (top three panels) containing either an m7GpppG or ApppG 5' cap structure or lacking a cap structure (pppG) or mutant U3D RNA (bottom panel) containing an m7GpppG cap were made in vitro and injected into nuclei of oocytes. At the times indicated, the nucleocytoplasmic distributions of the RNAs were analyzed as in (B). The m7G cap of wild-type U3 RNA (top panel) is nearly fully converted to the m2'2'7G cap within 6 h in the nucleus (Figure IB), whereas that of mutant U3D RNA (which lacks functional Box D sequences) is not (Figure SA). Lane 1 (M) shows the RNAs prior to injection.

U8 but not U3 RNA in the nucleus. It is possible that on the nature of its 5' cap structure (Figure 4C). For Box D promotes the retention of U3 RNA in the nucleus example, U3 RNA containing a non-physiological ApppG but its potential role remains unclear because U3 RNA cap structure was stable and remained in the nucleus apparently contains an additional nuclear retention ele- (second panel). Furthermore, a mutant U3 RNA that lacked ment. Box D sequences are also necessary for full stability a sequence element required for efficient hypermethylation of both U3 and U8 RNAs, but U8 RNA is much more (Box D sequences, see below) remained in the nucleus in affected by a change in this sequence (compare RNA an m7G-capped form (bottom panel). Some type of cap levels in Figure 4B, lane 1 with lanes 4 and 5). structure was necessary for stability since uncapped U3 Retention of U3 RNA in the nucleus does not depend RNA was rapidly degraded (third panel). 4863 M.P.Terns et aL

sF.4ss:...

I' .40 M.:'. Abiam, ... 9 ;m...

:. zL: ,.

.Iftoo

.ft:... :::.

Fig. 5. Identification of signals for hypermethylation and stability of U3 and U8 RNAs in the nucleus. (A) Requirement of Box D for Fig. 6. Nuclear retention and hypermethylation of the intron-encoded efficient hypermethylation of both U3 and U8 RNAs in the nucleus. U14 small nucleolar RNA. of U14 32P-labeled m7G-capped mutant U3 and U8 D (A) Hypermethylation m7G-capped (Box substitutions) and RNA in the nucleus. U14 RNAs made in vitro with (m7GpppG) or m7G-capped wild-type Ul and U2 RNAs (as controls) were co- without (pppG) an m7G cap were injected into nuclei of oocytes. After injected into nuclei of oocytes and 6 h later the nuclear RNAs were 6 h, the oocytes were fractionated and the RNAs present in nuclear isolated and immunoprecipitated with anti-m7G or anti-m2'2'7G cap samples were precipitated with anti-m7G or anti-m2'2'7G antibodies as antibodies; total nuclear RNA (T) and the RNAs purified from both indicated. Total nuclear RNA (T) and RNAs from both the precipitate the precipitate (P) and supernatant (S) fractions were analyzed by (P) and supernatant (S) fractions were analyzed by PAGE. PAGE. The low level of mutant U8 RNA (T, lane 1) results in part (B) Dependency of Box D for hypermethylation of m7G-capped U14 from nuclear instability of U8AD RNA and in part from its export to RNA in the nucleus. An m7G-capped mutant U14 RNA (U14AD) that the cytoplasm (Figure 4B, lane 5). (B) Requirement of an intact 3' lacked functional Box D sequences was co-injected with U1 and U6 terminal base-paired stem structure for efficient hypermethylation of RNAs into nuclei of oocytes. After 2 h, RNAs present in the nuclear U3 RNA in the nucleus. Wild-type U3 (lanes 1-3) or a mutant U3 fractions were precipitated with anti-m7G and anti-m2,2,7G antibodies RNA (lanes 4-6; U3Astem) with a disrupted 3' stem were co-injected and analyzed as in (A). The Ul RNA served as an intemal control for with Ul RNA into nuclei of oocytes. After 2 h, U3 and U1 RNAs the ability of the antibody to recognize m2'2'7G-capped RNA. present in the nuclear fractions were precipitated with anti-m 2,27G (C) Retention of U14 RNA in the nucleus. The nucleocytoplasmic antibodies and the RNAs from both the precipitate (P) and supematant distributions of uncapped (top panel) and m7G-capped (bottom panel) (S) fractions were analyzed as in (A). Hypermethylation of pre-Ul U14 RNAs after their into RNA in this injection nuclei of oocytes was determined experiment served as an internal control for the ability of as described in Figure IA. Lane 1 (M) shows the RNAs prior to antibodies to recognize m2'2'7G-capped RNAs. Lanes 1 and 4 show the injection. total (T) sample prior to fractionation by immunoprecipitation. (C) Requirement of an intact 3' terminal stem for stability of U3 RNA in the nucleus. The levels of mutant U3 RNA (U3Astem) in nuclei (N) and cytoplasm (C) were analyzed at 2 and 6 h following injection of (Leverette et al., 1992; reviewed in Sollner-Webb, 1993) the RNA into oocyte nuclei. and have 5' monophosphate termini which remain uncapped. We asked if such intron-encoded snoRNAs Structural elements needed for nuclear contained signals that would support their hypermethyl- hypermethylation of snoRNAs ation in the nucleus if they had m7G caps. An m7G-capped U3 and U8 RNAs lacking Box D sequences did not U14 RNA was hypermethylated in nuclei (Figure 6A, lane undergo significant hypermethylation of their 5' caps 4) in a Box D-dependent manner (Figure 6B, lane 4). (Figure 5A). Immunoprecipitation of the co-injected Ul Both uncapped and m7G-capped U14 RNA remained in and U2 RNAs using the m2'27cap antibodies showed that the nucleus (Figure 6C, lanes 2 and 4). the antibodies could have detected the modified RNAs if they had been appropriate substrates for hypermethylation Hypermethylation of snRNAs in nuclear extracts (lane 4). from Xenopus oocytes For U3 RNA, an intact 3' base paired stem was also To characterize the activity responsible for hypermethyl- necessary for its efficient nuclear hypermethylation even ation of snoRNAs in Xenopus oocyte nuclei, we developed when the Box D sequences were present (Figure SB, lanes an in vitro system using nuclear extracts and in vitro 2 and 5). Because other m2,27G-capped snoRNAs appear synthesized RNAs that contain a single 32P-label in the to lack similar 3' stem structures (Figure 4A; Reddy et al., m7G cap structure. After incubation, the RNA was purified 1985; Tyc and Steitz, 1989) a role of the stem structure and analyzed for cap modifications either by immuno- in hypermethylation appears to be specific to U3 RNA. precipitation with specific antibodies or by digestion with This 3' stem was also required for stability of U3 RNA nuclease P1 followed by thin layer chromatography (TLC) in the nucleus (Figure SC). of the products (Silberklang et al., 1979). Several snoRNAs, such as U14 RNA, are processed An activity present in nuclear extracts hypermethylated from intron sequences within protein-encoding mRNAs most of the m7GpppG cap of the input pre-U3 RNA 4864 Maturation of small nucleolar RNAs A B C

< 0 + X-

m e Xm - - No r. W- :J 3 D D;D:~: D ;:: D: Ab: m7G m2,2,7G I - + c T FP -Sl_P-S Al

A A s, m2,27GpppG =_.. :: nm2,2,7GpppG l"M AvalmikimmuL. .OEM.. MENIM- .. m2,2,7GpppG maw--,...... lwi40% * Awwk.l _ :v 0 ....uh,S.-w .ail&. #..w*.*..*_ m7GpppG I m7GpppG * 't.^|....

:.... :.#.:} _: origin _i: origin origin _ .ij...... 1 2 3 4 5 6 7 1 2 3 4 5 1 2 34 566 78 9

Fig. 7. Efficient hypermethylation of U3 RNA in nuclear extracts. In vitro made RNAs with a single 32P-label in the m7G cap were incubated in nuclear extracts at 19°C for 3 h. After purification, the RNAs were digested with nuclease P1 (either with or without prior immunoprecipitation with cap-specific antibodies) and the products were analyzed by TLC (the positions of origin, m7GpppG and m2'2'7GpppG dinucleotides are marked). The spot indicated by * likely corresponds to a 2'-0-methylated cap dinucleotide (m7GpppGm), since treatment of input RNA (e.g. A; lane 1) with tobacco acid pyrophosphatase (which releases the 5' cap monophosphate) produced a single labeled product (not shown). After incubation in the nuclear extract, a novel spot (A) was observed that appears to be the hypermethylated form of spot * (not shown). (A) Requirements for hypermethylation of U3 RNA in nuclear extracts. Input U3 RNA is shown in lane 1. The RNA was incubated in nuclear extract in the presence (lane 2) or absence (lane 3) of 0.1 mM SAM or with 0.5 mM S-adenosyl-homocysteine (SAH) (lane 4). Standard (lane 2) indicates the incubation conditions described in Materials and methods. Four mM EDTA and EGTA were added to the reaction analyzed in lane 5 and ATP and creatine phosphate (CP) were omitted from the reaction in lane 6. The products of incubation in heat inactivated extract (10' at 650C) are shown in lane 7. (B) Analysis of 5' caps of U3 RNA after incubation in a nuclear extract. U3 RNA was incubated under standard conditions in nuclear extract, purified and immunoprecipitated either with anti-m7G or anti-m2,2,7G antibodies as indicated. Total nuclear RNA (T) and RNAs isolated from the precipitate (P) and supernatant (S) fractions were analyzed as in (A). (C) Specificity of hypermethylation for snoRNAs. All RNAs shown were m7G-capped, prepared identically (Materials and methods) and incubated under standard conditions at 19°C in nuclear extract. Analysis of the cap was as in (A).

(Figure 7A, lane 1) to m2,2,7GpppG (lane 2). The identities complexes with Sm proteins. Likewise, nuclear hyper- of the spots seen on the TLC plates were confirmed by methylation may also occur only on RNPs, and the nuclear immunoprecipitiation of the RNAs prior to nuclease P1 extract might serve as a source of RNA binding proteins digestion (Figure 7B). The efficiency of the reaction that could interact with snoRNAs, but not spliceosomal decreased in the absence ofexogenously added S-adenosyl- RNAs. Thus, the lack of hypermethylation of Ul RNA in methionine (SAM) (Figure 7A, lane 3) and was abolished the nuclear extract could reflect the absence of available in the presence of S-adenosyl-homocysteine (lane 4; Sm proteins needed to form these complexes. Indeed, after Ueland, 1982) showing that the donor for the methyl incubation in nuclear extracts at 19°C, the Ul RNA was group is SAM. The methyl transferase activity apparently not precipitable by anti-Sm antibodies (Figure 8A, lanes 2 did not require divalent cations, since the addition of and 3). EDTA and EGTA to the reaction had no significant effect To test if the methyl transferase activity present in (lane 5); ATP and creatine phosphate (CP) stimulated the nuclei could hypermethylate spliceosomal RNAs that reaction only slightly (compare lanes 2 and 6) and heating are in RNP complexes, pre-Ul and U2 RNAs were the extract abolished the hypermethylating activity (lane 7). reconstituted in vitro into snRNPs (Figure 8B, lanes 1 and The methyl transferase activity present in nuclear 2; Sumpter et al., 1992). The RNAs of these reconstituted extracts had the same substrate specificity as was observed snRNPs were hypermethylated upon injection into isolated in the oocyte injection experiments. Wild-type U3 and U8 nuclei (lanes 3 and 4) under incubation conditions (19°C) RNAs were efficiently hypermethylated in the nuclear where the same RNAs injected as naked molecules were extracts incubated at 19°C (Figure 7C, lanes 2 and 5) but not hypermethylated (lanes 5 and 6; Tems and Dahlberg, mutant RNAs lacking Box D sequences (U3 and U8 1994). Moreover, when U1 RNA was incubated in nuclear RNAs) or the 3' stem structure (U3 RNA) were not extracts under conditions (30°C) that promoted the (lanes 3, 4 and 6). Spliceosomal Ul RNA was not exchange of Sm proteins from endogenous nuclear par- hypermethylated in nuclear extracts nor was m7G-capped ticles to the added U1 RNA (Figure 8A, lanes 4-5 and U6 RNA (lanes 7-9). B, lanes 2-3), only U1 RNA which contained the Sm Hypermethylation of spliceosomal snRNAs occurs in the protein binding site was efficiently hypermethylated cytoplasm only after the RNAs have been assembled into (Figure 8C, lanes 4 and 5). We propose that the same 4865 M.P.Terns et aL

" V3 io 0

NW i-I., W a -ppUj v '3 _ , ~~~I iI !~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

V2 11 Fig. 9. Hypermethylation of U3 RNA in nuclear extract in the absence IA of nucleoli. (A) Analysis of U3 RNA hypermethylation in nuclear fractions. Nuclear extract was fractionated by centrifugation into a pellet fraction (P) containing nucleoli and a supernatant fraction (S) devoid of nucleoli. m7G-capped U3 RNA was incubated in each Fig. 8. Nuclear hypermethylation of Sm-bound spliceosomal snRNAs. fraction or the unfractionated extract (T) and the cap structures were (A) Formation of Sm-precipitable Ul RNPs in nuclear extracts at analyzed as in Figure 7A. (B) Absence of nucleoli from the nuclear 30°C. Four fmol of 32P-labeled, m7G-capped U3 and UlA- RNA supematant fraction. Endogenous nucleolar U3 RNA was assayed by (which has a wild-type Sm protein binding site but lacks the binding Northern blot hybridization, using 2.5 nuclear equivalents of total site for the UIA protein) was incubated in nuclear extract at 19 or nuclear RNA (T) or RNAs present in the P and S fractions (see A) 30°C. After 3 h, Sm-bound RNAs were isolated by immuno- and a mixture of 32P-labeled antisense Ul, U2 and U3 RNAs as precipitation with anti-Sm antibodies and the RNAs from both probes. precipitate (P) and supernatant (S) fractions were analyzed by PAGE. (B) Hypermethylation of spliceosomal RNAs in reconstituted snRNPs. U1 and U2 snRNPs were reconstituted in vitro using purified snRNP for hypermethylation of the U3 snoRNA cap structure are proteins and in vitro made 32P-labeled m7G-capped Ul and U2 RNAs. present in the nucleoplasmic compartment of nuclei. The efficiency of reconstitution of snRNP core particles (Sumpter et al., 1992) was determined by immunoprecipitation of the RNAs with anti-Sm antibodies (lanes 1 and 2). Reconstituted snRNPs (lanes Discussion 3 and 4) or naked RNA (lanes 5 and 6) were injected into oil-isolated nuclei and incubated at 18°C for 4 h. Hypermethylation was assayed In our analysis of the metabolism of snoRNAs we show by immunoprecipitation with anti-m 2,27G antibodies and the RNAs that the snoRNAs U3, U8 and U14 are retained from both precipitates (P) and supernatants (S) were analyzed as in selectively (A). (C) Requirement of an Sm protein binding site for hyper- and matured in the nucleus by mechanisms that are methylation of Ul RNA in nuclear extracts. One to two fmol of saturable and sequence-specific. A conserved sequence 32P-labeled m7G-capped U3, UlSm- (which lacks a functional Sm element Box D, present in most snoRNAs, plays a key protein binding site) and U1A- RNAs were incubated in nuclear extract role in the 5' cap hypermethylation, retention and stability for 3 h at 30°C prior to immunoprecipitation with anti-Sm or anti- m2'2'7G antibodies. The RNAs prepared from precipitates (P) and of these RNAs in the nucleus. We propose that substrates supematants (S) were analyzed as in (A). for nuclear hypermethylation are m7G-capped RNAs associated with specific proteins such as those that bind Box D. nuclear activity hypermethylates both nucleolar and spliceosomal capped snRNAs provided they are in specific Maturation pathway for snoRNAs RNP structures. These results also suggest that RNA Our findings that U3 RNA remains in the oocyte nucleus sequences such as Box D, which are required for nuclear where its 5' cap is hypermethylated (Terns and Dahlberg, hypermethylation of snoRNAs, are important for binding 1994) has recently been confirmed in mammalian cells the needed snoRNA-specific proteins. (Cheng et al., 1995). The fact that U8 RNA and U14 To determine if hypermethylation of snoRNAs occurred RNA behave analogously to U3 RNA (Figures 1 and 6) outside the nucleolus, nuclei were separated into two frac- indicates that most or all snoRNAs remain in nuclei tions which either contained or lacked nucleoli (Peculis and during their maturation. Precursor snoRNAs apparently Gall, 1992) and these fractions were tested for activity on are matured in the nucleoplasm before they are transported U3 RNA. The cap of U3 RNA was hypermethylated in the to the nucleolus, as the 5' caps of these snoRNAs are absence of nucleoli (Figure 9A, lane 2) and little or no hypermethylated in the absence of nucleoli (Figure 9). detectable hypermethylation of U3 RNA was observed The determinants that target mature snoRNAs to nucleoli when the RNA was incubated in the nucleolar fraction (lane remain to be identified. 1). Mixing the two fractions resulted in a modest stimulation over the nucleoplasmic fraction alone, indicating that a Evidence for an active retention mechanism that stimulatory activity may be associated with the nucleolar keeps snoRNAs in the nucleus fraction (lane 3). The absence ofendogenous U3 RNA from The appearance of U3 and U8 RNAs in the cytoplasm the nucleoplasmic fraction showed that this fraction was only when their nuclear levels are elevated (Figures 2 and free of nucleoli (Figure 9B). Thus, the activities required 3) indicates that they are normally retained in the nucleus. 4866 Maturation of small nucleolar RNAs

The specificity of this retention is demonstrated by the (Jarmolowski et al., 1990; Huang et al., 1992) and ability of U3 or U8, but not Ul RNA (Figure 3) or U6 U8 RNA in Xenopus oocytes (Figure 4B; Peculis and RNA (E.Lund, unpublished results), to saturate it. Oocyte Steitz, 1994). nuclei are capable of retaining <25 fmol of U3 RNA Our finding that m7G-capped U3 RNA remained in the (Figure 3 and data not shown). In contrast, as much as nucleus (Figure 4) further supports our previous results 600 fmol of U6 RNA can be injected into the nucleus that an m7G cap itself is not sufficient to direct RNA without leading to the appearance of U6 RNA in the transcripts out of the nucleus (Terns et al., 1993a). The cytoplasm (E.Lund, unpublished results). These findings m7G cap has been shown to facilitate the export of other indicate that both the number and type of nuclear retention snRNAs and mRNAs (Hamm and Mattaj, 1990; Terns sites differ for different RNAs. et al., 1993a; Jarmolowski et al., 1994) and its function A previous report concluded that U3 RNA is normally in export presumably is mediated by m7G cap binding exported to the cytoplasm (Baserga et al., 1992) since proteins (Izaurralde et al., 1992). It is unclear whether some U3 RNA appeared in the cytoplasm after injection specific m7G cap binding proteins interact with pre- of rat U3 genes into Xenopus oocyte nuclei. However, the snoRNAs and whether such cap binding proteins differ level of injected U3 genes in that earlier study was from those that recognize nucleoplasmic snRNAs and high, making it likely that U3 RNA accumulation in the mRNAs to promote their export. A trimethylguanosine cytoplasm resulted from saturation of the U3 RNA nuclear cap is not necessary for nuclear retention of U3 RNA retention sites rather than as a normal step in U3 RNA (Figure 4). metabolism. The subcellular distribution of U3 RNA has been The nature of the saturable nuclear retention sites for reported to change as myoblasts differentiate into myo- snoRNAs remains unclear. Our finding that U3 RNA tubules, with a significant proportion of the U3 RNA competes for the nuclear retention of U8 RNA (as well being detected in cytoplasmic fractions of differentiated as itself; Figure 3) indicates that a titratable RNA binding myotubules (Glibetic et al., 1992). Since this differentia- factor(s) that associates with both U3 and U8 RNAs is tion is also accompanied by a dramatic reduction in responsible for retaining these RNAs in the nucleus prior ribosome production, changes in the capacity of nucleoli to their incorporation into nucleoli. It is likely that nucleoli to bind U3 RNA might account for the loss of U3 RNA contribute to the retention of mature snoRNAs in the retention in the nucleolus. This effect may be indirect, nucleus, perhaps through specific interactions of snoRNAs resulting from leakage of nucleoplasmic U3 RNA into the with rRNA or specific nucleolar proteins. cytoplasmic fraction during aqueous cell fractionation. Several lines of evidence indicate that nuclear retention Alternatively, the levels or activity of U3 RNA binding of the snoRNAs depends on a protein that recognizes Box factor(s) responsible for retaining U3 RNA in the nucleolus D sequences. First, Box D sequences are protected from and nucleus may decline during differentiation. nuclease digestion of U3 snRNPs; this protection is observed in extracts of both mammalian cells (Parker and 5' cap hypermethylation in the nucleus Steitz, 1987) and trypanosomes (Hartshorne and Agabian, The Box D element is required for hypermethylation of 1994) indicating that Box D is a phylogenetically con- both U3 and U8 RNAs in the oocyte nucleus (Figure 5). served protein binding site. Second, a variant U8 RNA It is interesting that U14 snoRNA (which normally lacks lacking Box D accumulates in the cytoplasm following a cap structure) is efficiently hypermethylated when it injection into the nucleus (Figure 4B). Finally, increasing contains an m7G cap, provided the RNA contains a Box the levels of wild-type U3 and U8 RNAs leads to trimming D element (Figure 6). Other snoRNAs which contain Box of these RNAs at their 3' ends (Figure 3A), suggesting D sequences also undergo hypermethylation when they that a protein that recognizes sequences near the 3' termini have a methylatable m7G cap. These include plant U3 of both of these RNAs, such as Box D, is limiting (Figure RNA (which normally contains a methyl-phosphate cap 4A). Thus, the putative Box D binding protein may serve structure, Shimba et al., 1992) when it is expressed from both to anchor the RNA to nuclear or nucleolar structures an RNAP II promoter (Kiss et al., 1991) and yeast U14 and to protect snoRNAs against degradation by nucleases. snoRNA (which normally lacks a cap structure) when it As shown in Figures 5-7, Box D is needed for hyper- is expressed from a yeast GALl (RNAP II) promoter methylation of the 5' caps of the snoRNAs, indicating (Balakin et al., 1994). In contrast, U6 RNA, an snRNA that a Box D binding protein may also interact with the that lacks Box D, is not hypermethylated even when it nuclear methyl transferase. has an m7G cap and is retained in the nucleus (Terns The requirement for Box D in the retention of U8 but et al., 1993a). These examples also indicate that nuclear not U3 RNA (Figure 4B) indicates that U3 RNA contains hypermethylation operates in the cells of organisms as another nuclear retention signal(s). This idea is supported divergent as yeast, plants and vertebrates. by the fact that retention of U3 RNA can be efficiently In addition to Box D, an intact 3' base-paired stem was competed by U3 RNA but not by U8 RNA, whereas required for hypermethylation of U3 RNA within the retention ofU8 RNA is competed, with similar efficiencies, nucleus (Figure 5B). Previously, it was reported that by U3 or U8 RNA (Figure 3A). Nuclear retention of U14 disruption of the 3' terminal stem structure resulted in RNA may also require Box D sequences, since a fraction complete loss of hypermethylation of U3 RNA injected of U 14 RNA lacking functional Box D sequences appeared in the oocyte cytoplasm but that substitution of box D in the cytoplasm (data not shown); however, the extreme sequences decreased hypermethylation only 2- to 3-fold instability of this mutant U14 RNA in nuclei complicates (Baserga et al., 1992). Comparable 3' stems are apparently interpretation of these results. Box D was also found to absent from the other snoRNAs that become hyper- be important for the stability of U14 RNA in yeast methylated (Reddy et al., 1985; Tyc and Steitz, 1989) 4867 M.P.Terns et aL indicating that in U3 RNA, this structure is needed to This raises the possibility that snoRNAs may associate with stabilize the binding of proteins to the neighboring Box fibrillarin indirectly through protein-protein interactions. D sequences. Other sequences or structures that flank Box Recently, three U3 RNA binding proteins, with molecular D in other snoRNAs are also likely to be important for weights of -15, 50 and 55 kDa, were purified from CHO protein binding and RNA hypermethylation. cell extracts (Lubben et al., 1993). Unlike fibrillarin, An intact 3' terminal stem is also critical for the stability which exhibited a salt-sensitive interaction and did not of U3 RNA in the nucleus (Figure 6C). Thus, the 3' stem remain associated with U3 RNA during this purification may not be required for U3 RNA import as previously procedure, these three tightly associated proteins may be reported (Baserga et al., 1992) but rather for stabilization 'core' U3 RNA binding proteins. The 55 kDa protein of U3 RNA once it has been imported to the nucleus. A interacted with the central region of U3 RNA and is similar 3' terminal base-paired sequence is also required therefore unlikely to interact with Box D at the 3' terminus for the stability of U14 in yeast (Huang et al., 1992). of U3 RNA. Also, using antibodies against fibrillarin, six Surprisingly, m7G-capped spliceosomal snRNAs can be other candidate snoRNA proteins have been identified in hypermethylated in nuclei if the RNAs are complexed mammalian cells (Parker and Steitz, 1987) and a U3- with Sm proteins (Figure 8). Previously, this was believed specific protein has been identified in yeast (Jansen et al., to be exclusively a cytoplasmic event. Nuclear hyper- 1993). The binding sites in U3 RNA for these proteins methylation may have gone unnoticed since pre-snRNAs have not been determined. normally are rapidly hypermethylated in the cytoplasm It is likely that RNA binding proteins will play important soon after they associate with free Sm proteins (Mattaj, roles in several aspects of snoRNA metabolism including 1986). The trimethylguanosine cap is necessary for the the maturation, nucleolar localization and function of the import of only certain snRNA species (Fischer et al., RNAs. By analogy with spliceosomal snRNAs, which 1991) and for these snRNAs, only in certain cell types contain both common 'core' proteins (the Sm proteins) (Fischer et al., 1994; Marshallsay and Luhrmann, 1994). and snRNA specific proteins, we expect that snoRNA Thus, some spliceosomal snRNAs may escape hyper- binding proteins will fall into two classes: those that methylation in the cytoplasm and become modified in the interact with common sequence elements and those that nucleus following import. In either case, hypermethylation interact with sequences unique to the many different in the cytoplasm (Plessel et al., 1994) or nucleus appears snoRNAs. to occur only to RNAs that are complexed with specific snRNP proteins. It is conceivable that the 5' caps of mature nuclear Materials and methods snRNAs may undergo reversible changes in their methyl- ation status, perhaps while they function in pre-RNA DNA template generation and in vitro transcription In Templates encoding either wild-type or mutant snRNAs used for in vitro processing. that case, enzymes capable of both adding synthesis of snRNAs were either linearized plasmid (containing and removing specific methyl groups from the caps of SP6 or T7 phage RNA promoter sequences) or DNA fragments generated these snRNAs would be expected to exist in nuclei. Further by PCR amplification of the RNA coding regions of the plasmids using studies will test this possibility and will show whether appropriate 5' and 3' primer pairs (as described below). The 5' primer the same nuclear acts on both used in the PCR reactions contained either SP6 or T7 polymerase hypermethylase activity promoter sequences. All PCR-derived template DNA was confirmed by spliceosomal and snoRNAs. The development of the DNA sequencing using the dideoxynucleotide chain termination method in vitro system for nuclear hypermethylation described using Sequenase or Cycle sequencing kits (USB). here will facilitate the identification and purification of the Plasmids containing snRNA genes have been described previously nuclear enzyme(s) containing hypermethylating activity. and include: Xenopus laevis Ul (pXlUlbl; Krol et al., 1985), X.laevis U2 (pXIU2.5; Mattaj and De Robertis, 1985), Xlaevis U3A (pXIU3A'; Savino et al., 1992), Xenopus tropicales U6 gene (pXTU6-2; Krol et al., snoRNA binding proteins 1987), X.laevis U8 (pSPU8.2; Peculis and Steitz, 1993) and mouse It is very likely that the function of Box D in several U14.5 (pBT20; Shanab and Maxwell, 1992). aspects of snoRNA metabolism is mediated through its PCR templates used to transcribe U1 and U6 RNAs were as previously interaction with a specific RNA binding protein(s) in described (Tems et al., 1993a). The U3 template differed from the one the nucleus. Proteins that used previously (Terns and Dahlberg, 1994) in that it contained a T at specifically interact with the nucleotide position 220 (Savino et al., 1992). The resultant U3 RNA is conserved Box D element and other sequences of snoRNAs one nucleotide longer than the reported mature U3 RNA (Jeppesen et al., remain to be identified and characterized. Fibrillarin is the 1988) and was included since the encoded A-220 is predicted to form a only protein known to be common to different snoRNPs five base-paired 3' terminal stem which is phylogenetically conserved but several observations indicate that fibrillarin may not (Parker and Steitz, 1987). be To make U2 DNA template, pXIU2.5 (Mattaj and De Robertis, 1985) responsible for the activities of Box D described was amplified using a 5' primer consisting of SP6 promoter sequences in this work. For example, mutations that prevent the and sequences complementary to U2 nucleotides 1-16 and a 3' primer association of fibrillarin to U3 RNA both in vitro (Baserga which is complementary to the 3' terminal 14 nucleotides of mature U2 et al., 1991) and in vivo (Baserga et al., 1992) do not RNA plus five additional encoded nucleotides, so that transcribed U2 affect hypermethylation of U3 RNA injected into Xenopus RNA resembles the 3'-extended precursor U2 RNA (Neuman de Vegvar and Dahlberg, 1990). Templates for Xenopus U8 (Peculis and Steitz, oocytes (Baserga et al., 1992). Also, the stability, 5' 1993) and mouse U14.5 (Shanab and Maxwell, 1992) RNAs were cap hypermethylation or targeting of many snoRNAs to linearized plasmids. nucleoli in yeast cells are unaffected by depletion of Mutagenesis of the Box D sequences of U3, U8 and U14 genes was fibrillarin (Tollervey et al., 1991, 1993). performed by PCR amplification ofthe genes using mutagenic 3' primers. Base substitutions most snoRNAs are anti- introduced in the positon of the Box D were TTITTT Although co-precipitated by (wild-type is GGCTGA) for U3, GTCTAG (wild-type is TTCTGA) for fibrillarin antibodies, a direct interaction of this protein U8 and TCTAGA (wild-type is GTCTGA) for U14 RNA. To disrupt with any of the snoRNAs has not yet been demonstrated. terminal stem formation in the U3 RNA, the 3' terminal nucleotides of 4868 Maturation of small nucleolar RNAs

U3 gene (GTGGT) were substituted with CACCC. Sequences mutated After incubation for 1 h at 37°C, the RNA was gel purified and stored in U3 were identical to those of a previous study (Baserga et al., 1992) in H20 at -20°C. In vitro transcription of T7 or SP6 DNA templates was performed as previously described (Terns et al., 1993a) and all RNA transcripts were Cap hypermethylation in nuclear extract and analysis of the purified after electrophoresis in 8% denaturing polyacrylamide gels. cap structure The standard reaction contained 4 pd of nuclear extract (4 nuclear Oocyte micro-injection analysis of RNA transport equivalents) incubated for 3 h at 19°C with 1 pd of a mixture containing Stage V and VI oocytes, obtained from X.laevis frogs, were injected 1.6 mg/ml tRNA, 4 mM ATP, 100 mM CP, 0.5 mM SAM and 0.1- into their nuclei with 12 nl of solution. The injection solution contained 0.2 fmoll of cap-labeled RNA substrate (dependent on the efficiency -1 fmol of each 32P-labeled in vitro-made snRNA (for direct RNA of the capping reaction, this mixture contained also -1-5 fmol/l of injections) or snRNA genes (to synthesize snRNAs in vivo; amount of unlabeled, uncapped RNA which copurified with the labeled RNA). DNA is indicated in figure legends). For the gene injection experiments, After incubation, the reaction mixture was either used directly for DNA (in the presence or absence of antibodies; see Figure 2) was immunoprecipitation using anti-Sm antibodies or it was incubated with injected into nuclei and 1 h later the same oocytes received a second 80 pl proteinase K (1 mg/ml) for 30 min at 37°C. The RNA was cytoplasmic injection of [a-32P]GTP (1 gCi/oocyte) to radiolabel the prepared by phenol-chloroform-isoamylalcohol (25:24:1) extraction and transcribed snRNAs in vivo. In Figure 8C, injections were done in oil- ethanol precipitation. The pelleted RNA was dissolved in 2 pd H20. A isolated nuclei (devoid of cytoplasm) as described previously (Terns and 0.5 pl sample was run on a denaturing polyacrylamide gel to assay for Dahlberg, 1994). To monitor the accuracy of the nuclear injections, stability of the RNA after incubation in the extract, and the remainder injection mixtures contained both blue dextrans which are too large to (1.5 pl) was incubated with 1 pd of nuclease P1 (1 U/pd) for 1 h at diffuse from the nucleus and allows a direct visualization of the injection 37°C. The P1 digestion products were spotted directly on a thin layer site (Jarmolowski et al., 1994) and 32P-labeled U6 RNA, since this RNA cellulose chromatography plate and resolved in isobutyric acid/NH40H/ remains within the nucleus (Hamm and Mattaj, 1989; Terns et al., 1993a). H20 (66:1:33); the products were detected by autoradiography of At different times after the injections, oocytes were separated into the plates. nuclear and cytoplasmic fractions by manual dissection under mineral oil (Lund and Paine, 1990) and pools of 3-4 oocytes containing 'blue' nuclei were further analyzed. RNAs were purified and analyzed by Acknowledgements electrophoresis in 8% polyacrylamide gels (containing 7 M urea) and autoradiography as described (Terns et al., 1992). We thank Ariane Grandjean and Michele Barr for technical assistance. We wish also to thank Iain Mattaj, Susan Gerbi, Alain Krol, Brenda Immunoprecipitation Peculis, Joan Steitz and Stuart Maxwell for kindly providing snRNA Extracts were prepared from isolated nuclei and immunoprecipitations genes, Utz Fischer and Reinhard Luhrmann for supplying purified snRNP were performed as previously described using IpplSO buffer (Terns proteins and Ram Reddy for furnishing the y-mpppG cap analogue. Also et al., 1992). The antibodies used were rabbit polyclonal antibodies we thank Theodore Munns, Reinhard Luhrmann, K.Michael Pollard, against the m2'27G (Bringmann et al., 1983) (from R.Luhrmann) and Eng Tan, David Goldfarb and Joan Steitz for providing antibodies used m7G cap (Munns et al., 1982) (from T.Munns); and mouse monoclonal in this study. This work was supported by NIH grant GM30220 to J.E.D. antibodies against Sm proteins (Lerner et al., 1981) (mAb Y12 from and postdoctoral fellowships from the Swiss National Science Foundation J.Steitz) and fibrillarin (Reimer et al., 1987) (mAb 72B9 from K.M. to C.G. and from NIH (GM14704) and ACS (PF4121) to M.P.T. Pollard and E.Tan). Deproteinized RNAs were used for precipitation with the anti-cap antibodies and extracts were used to co-precipitate the RNAs with anti-Sm protein and anti-fibrillarin antibodies. References Preparation and fractionation of nuclear extract Balakin,A.G., Lempicki,R.A., Huang,G.M. and Fournier,M.J. (1994) Stage VI oocytes were manually dissected under mineral oil as described Saccharomyces cerevisiae U14 small nuclear RNA has little secondary (Lund and Paine, 1990) and nuclei were collected in a 0.5 ml microfuge structure and appears to be produced by post-transcriptional processing. tube on ice. After a 4 s spin at 4°C, the excess oil over the nuclear J. Biol. Chem., 269, 739-746. pellet was removed. Nuclei were homogenized in 1 nucleus/pul ice cold Baserga,S.J., Gilmore,H.M. and Yang,X.W. (1992) Distinct molecular buffer D [50 mM Tris-HCl (pH 7.6), 25 mM KCI, 5 mM MgCl2, 3 mM signals for nuclear import of the nucleolar snRNA, U3. Genes Dev., DTT, 250 mM sucrose] by repeated pipetting (-10 strokes) with a Rainin 6, 1120-1130. P-200 (yellow) pipette tip. The homogenate was centrifugated for 3 min Baserga,S.J. and Steitz,J.A. (1993) The diverse world of small at 10 000 r.p.m. at 4°C and the cleared supernatant collected. After the ribonucleoproteins. In Gesteland,R.F. and Atkins,J.F. (eds), The RNA addition of 1 U/gl RNAsin (Promega), 10 ng/ml aprotinin, 1 ng/ml each World. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, of leupeptin and pepstatin, the crude nuclear extract was quick frozen NY, pp. 359-381. and stored at -70°C. Typical protein-concentration of the extract was Baserga,S.J., Yang,X.D. and Steitz,J.A. (1991) An intact Box C sequence -2 mg/ml. in the U3 snRNA is required for binding of fibrillarin, the protein Fractionation of the nuclear extract was carried out by homogenizing common to the major family of nucleolar snRNPs. EMBO J., 10, oil-isolated nuclei in modified buffer D containing 10 mM MgCl2. To 2645-2651. pellet the nucleoli, the homogenate was centrifuged for 30 min at 14 000 Birnstiel,M.L. and Schaufele,F.J. (1988) Functions of the abundant U- r.p.m. at 4°C. The supernatant served as a nucleoplasmic fraction. The snRNPs. In Birnstiel,M.L. (ed.), Structure and Function ofMajor and pellet was washed once with cold modified buffer D, centrifuged again Minor Small Nuclear Ribonucleoprotein Particles. Springer-Verlag, and resuspended at 1 nucleus/pl modified buffer D. RNAsin and protease Berlin, pp. 155-182. inhibitors were added to both fractions as for crude nuclear extract. Bringmann,P., Rinke,J., Appel,B., Reuter,R. and Luhrmann,R. (1983) Northern blot analysis was carried out using 2.5 nuclear equivalents of Purification of snRNPs Ul, U2, U4, U5 and U6 with 2,2,7- RNA from total extract and both the pellet and supernatant fractions. trimethylguanosine-specific antibody and definition of their constiuent The RNAs were separated by denaturing PAGE and transferred onto a proteins reacting with anti-Sm and anti-(U1) RNP antisera. EMBO J., nylon membrane using a semi-dry blotting system (Bio-Rad) according 2, 1129-1135. to manufacturers recommendations and the membranes were probed Caizergues-Ferrer,M., Mathieu,C., Mariottini,P., Amalric,F. and with 32P-labeled anti-sense snRNAs. Amaldi,F. (1991) Developmental expression of fibrillarin and U3 snRNA in Xenopus laevis. Development, 112, 317-326. In vitro capping reaction with guanylyltransferase from Cheng,Y., Dahlberg,J.E. and Lund,E. (1995) Diverse effects of the vaccinia virus guanine nucleotide exchange factor RCC1 on RNA transport. Science, The substrates for the in vitro capping reaction were unlabeled gel 267, 1807-1810. purified RNAs with 5' triphosphate ends transcribed by SP6 or T7 Davis,L.I. and Blobel,G. (1986) Identification and Characterization of a polymerases in vitro using Promega's Ribomax transcription conditions. Nuclear Pore Complex Protein. Cell, 45, 699-709. The capping reaction (Shuman and Moss, 1990) contained 50 mM Tris- Filipowicz,W. and Kiss,T. (1993) Structure and function of nucleolar HCl (pH 7.8), 1.25 mM MgCI2, 6 mM KCl, 2.5 mM DYT, 50 jiM SAM, snRNPs. Mol. Biol. Rep., 18, 149-156. 0.1 mg/ml BSA, 2 U/pl RNasin, 0.3 ,ug RNA, 100 ,uCi [a-32P]GTP Fischer,U., Darzynkiewicz,E., Tahara,S.M., Dathan,N.A. and (800 Ci/mmol) and 4 U vaccinia virus guanylytransferase (Gibco, BRL). Luhrmann,R. (1991) Diversity in the signals required for the nuclear 4869 M.P.Terns et aL

accumulation of U snRNPs and variety in the pathways of nuclear Li,H.D., Zagorski,J. and Fournier,M.J. (1990) Depletion of U14 small transport. J. Cell Biol., 113, 705-714. nuclear RNA (snR128) disrupts production of 18S rRNA in Fischer,U., Heinrich,J., van Zee,K., Fanning,E. and Luhrmann,R. (1994) Saccharomyces cerevisiae. Mol. Cell. Biol., 10, 1145-1152. Nuclear transport of Ul snRNP in somatic cells: differences in signal Lubben,B., Marshallsay,C., Rottmann,N. and Luhrmann,R. (1993) requirement compared with Xenopus laevis oocytes. J. Cell Biol., 125, Isolation of U3 snoRNP from CHO cells: a novel 55 kDa protein 971-980. binds to the central part of U3 snoRNA. Nucleic Acids Res., 21, Fischer,U. and Luhrmann,R. (1990) An essential signalling role for the 5377-5385. m3G cap in the transport of Ul snRNP to the nucleus. Science, 249, Ltihrmann,R., Kastner,B. and Bach,M. (1990) Structure of spliceosomal 786-790. snRNPs and their role in pre-mRNA splicing. Biochim. Biophys. Acta, Fischer,U., Sumpter,V., Sekine,M., Satoh,T. and Luhrmann,R. (1993) 1087, 265-292. Nucleo-cytoplasmic transport of U snRNPs: definition of a nuclear Lund,E. and Paine,P.L. (1990) Nonaqueous isolation of transcriptionally localization signal in the Sm core domain that binds a transport active nuclei from Xenopus oocytes. Methods Enzymol., 181, 36-43. receptor independently of the m3G cap. EMBO J., 12, 573-583. Marshallsay,C. and Luhrmann,R. (1994) In vitro nuclear import of Fournier,M.J. and Maxwell,E.S. (1993) The nucleolar snRNAs: catching snRNPs: cytosolic factors mediate m3G-cap dependence of Ul and up with the spliceosomal snRNAs. Trends Biochem. Sci., 18, 131-135. U2 snRNP transport. EMBO J., 13, 222-231. Glibetic,M., Larson,D.E., Sienna,N., Bachellerie,J.P. and Sells,B.H. Mattaj,I.W. (1986) Cap trimethylation of U snRNA is cytoplasmic and (1992) Regulation of U3 snRNA expression during myoblast dependent on U snRNP protein binding. Cell, 46, 905-911. differentiation. Exp. Cell Res., 202, 183-189. Mattaj,I.W. and De Robertis,E.M. (1985) Nuclear segregation of U2 Green,M.R. (1991) Biochemical mechanisms of constitutive and snRNA requires binding of specific snRNP proteins. Cell, 40, 111-118. regulated pre-mRNA splicing. Annu. Rev. Cell Biol., 7, 559-599. Maxwell,E.S. and Fournier,M.J. (1995) The small nucleolar RNAs. Guthrie,C. and Patterson,B. (1988) Spliceosomal snRNAs. Annu. Rev. Annu. Rev. Biochem., 35, 897-933. Genet., 22, 387-419. Morrissey,J.P. and Tollervey,D. (1995) Birth of the snoRNPs: the Hamm,J., Darzynkiewicz,E., Tahara,S.M. and Mattaj,I.W. (1990) The evolution of RNase MRP and the eukaryotic pre-rRNA-processing trimethylguanosine cap structure of Ul snRNA is a component of a system. Trends Biochem. Sci., 20, 78-82. bipartite nuclear targeting signal. Cell, 62, 569-577. Munns,T.W., Liszewski,M.K., Tellam,J.T., Sims,H.F. and Rhoads,R.E. Hamin,J. and Mattaj,I.W. (1989) An abundant U6 snRNP found in germ (1982) Antibody-nucleic acid complexes. Immunospecific retention cells and embryos of Xenopus laevis. EMBO J., 8, 4179-4187. of globin messenger ribonucleic acid with antibodies specific for 7- Hamm,J. and Mattaj,I.W. (1990) Monomethylated cap structures facilitate methylguanosine. Biochemistry, 21, 2922-2928. RNA export from the nucleus. Cell, 63, 109-118. Neuman de Vegvar,H. and Dahlberg,J.E. (1990) Nucleocytoplasmic Hartshorne,T. and Agabian,N. (1994) A common core structure for U3 transport and processing of small nuclear RNA precursors. Mol. Cell. small nucleolar RNAs. Nucleic Acids Res., 22, 3354-3364. Biol., 10, 3365-3375. Huang,G.M., Jarmolowski,A., Struck,J.C. and Fournier,M.J. (1992) Parker,K.A. and Steitz,J.A. (1987) Structural analysis of the human U3 Accumulation of U14 small nuclear RNA in Saccharomyces cerevisiae ribonucleoprotein particle reveal a conserved sequence available for requires box C, box D, and a 5', 3' terminal stem. Mol. Cell. Biol., base pairing with pre-rRNA. Mol. Cell. Biol., 7, 2899-2913. 12, 4456-4463. Peculis,B.A. and Gall,J.G. (1992) Localization of the nucleolar protein Hughes,J.M. and Ares,M.,Jr (1991) Depletion of U3 small nucleolar No38 in amphibian oocytes. J. Cell Biol., 116, 1-14. RNA inhibits cleavage in the 5' external transcribed spacer of yeast Peculis,B.A. and Steitz,J.A. (1993) Disruption of U8 nucleolar snRNA pre-ribosomal RNA and impairs formation of 18S ribosomal RNA. inhibits 5.8S and 28S rRNA processing in the Xenopus oocyte. Cell, EMBO J., 10, 4231-4239. 73, 1233-1245. Izaurralde,E., Stepinski,J., Darzynkiewicz,E. and Mattaj,I.W. (1992) A Peculis,B.A. and Steitz,J.A. (1994) Sequence and structural elements cap binding protein that may mediate nuclear export of RNA critical for U8 snRNP function in Xenopus oocytes are evolutionarily polymerase II-transcribed RNAs. J. Cell Biol., 118, 1287-1295. conserved. Genes Dev., 8, 2241-2255. Jansen,R., Tollervey,D. and Hurt,E.C. (1993) A U3 snoRNP protein with Plessel,G., Fischer,U. and LUhrmann,R. (1994) m3G Cap hyper- homology to splicing factor PRP4 and Go domains is required for methylation of Ul small nuclear ribonucleoprotein (snRNP) in vitro: ribosomal RNA processing. EMBO J., 12, 2549-2558. evidence that the Ul small nuclear RNA-(guanosine-N2)-methyl- Jarmolowski,A., Boelens,W.C., Izaurralde,E. and Mattaj,I.W. (1994) transferase is a non-snRNP cytoplasmic protein that requires a binding Nuclear export of different classes of RNA is mediated by specific site on the sm core domain. Mol. Cell. Biol., 14, 4160-4172. factors. J. Cell Biol., 124, 627-635. Reddy,R., Henning,D. and Busch,H. (1985) Primary and secondary Jarmolowski,A., Zagorski,J., Li,H.V. and Fournier,M.J. (1990) structure of U8 small nuclear RNA. J. Biol. Chem., 260, 10930-10935. Identification of essential elements in U14 RNA of Saccharomyces Reimer,G., Pollard,K.M., Penning,C.A., Ochs,R.L., Lishwe,M.A., cerevisiae. EMBO J., 9, 4503-4509. Busch,H. and Tan,E.M. (1987) Monoclonal autoantibody from a (New Jeppesen,C., Stebbins-Boaz,B. and Gerbi,S.A. (1988) Nucleotide white) Fl mouse and some human of Xenopus U3 Zealand blackXNew Zealand sequence determiniation and secondary structure scleroderma sera target an Mr 34,000 nucleolar protein of the U3 snRNA. Nucleic Acids Res., 16, 2127-2148. RNP particle. Arthritis Rheum., 30, 793-800. Kass,S., Tyc,K., Steitz,J.A. and Sollner-Webb,B. (1990) The U3 small of Xenopus U3 functions in the first step of preribosomal Savino,R. and Gerbi,S.A. (1990) In vivo disruption nucleolar ribonucleoprotein snRNA affects ribosomal RNA processing. EMBO J., 9, 2299-2308. RNA processing. Cell, 60, 897-908. (1992) Genes for Xenopus laevis U3 Kiss,T. and Filipowicz,W. (1993) Small nucleolar RNAs encoded by Savino,R., Hitti,Y. and Gerbi,S.A. gene RCC1. EMBO J., 12, small nuclear RNA. Nucleic Acids Res., 20, 5435-5442. introns of the human cell cycle regulatory of 2913-2920. Shanab,G.M. and Maxwell,E.S. (1991) Proposed secondary structure Kiss,T., Marshallsay,C. and Filipowicz W. (1991) Alteration of the RNA eukaryotic U14 snRNA. Nucleic Acids Res., 19, 4891-4894. polymerase specificity of U3 snRNA genes during evolution and Shanab,G.M. and Maxwell,E.S. (1992) Determination of the nucleotide in vitro. Cell, 65, 517-526. sequences in mouse U14 small nuclear RNA and 18S ribosomal RNA Krol,A., Carbon,P., Ebel,J.P. and Appel,B. (1987) Xenopus tropicalis U6 responsible for in vitro intermolecular base-pairing. Eur J. Biochem., snRNA genes transcribed by Pol III contain the upstream promoter 206, 391-400. elements used by Pol II dependent U snRNA genes. Nucleic Acids Shimba,S., Buckley,B., Reddy,R., Kiss,T. and Filipowicz,W. (1992) Cap Res., 15, 2463-2478. structure of U3 small nucleolar RNA in animal and plant cells is Krol,A., Lund,E. and Dahlberg,J.E. (1985) The two embryonic Ul different. gamma-monomethyl phosphate cap structure in plant RNA. RNA genes of Xenopus laevis have both common and gene-specific J. Biol. Chem., 267, 13772-13777. transcription signals. EMBO J., 4, 1529-1535. Shuman,S. and Moss,B. (1990) Purification and use of vaccinia virus Lerner,E.A., Lerner,M.R., Janeway,C.A.,Jr and Steitz,J.A. (1981) messenger RNA capping enzyme. Methods Enzymol., 181, 170-180. Monoclonal antibodies to nucleic acid-containing cellular constituents: Silberklang,M., Gillum,A.M. and RajBhandary,U.L. (1979) Use of Probes for molecular biology and autoimmune disease. Proc. Natl in vitro 32p labeling in the sequence analysis of nonradioactive tRNAs. Acad. Sci. USA, 78, 2737-2741. Methods Enzymol., 59, 58-109. Leverette,R_D., Andrews,M.T. and Maxwell,E.S. (1992) Mouse U14 Sollner Webb,B. (1993) Novel intron-encoded small nucleolar RNAs. snRNA is a processed intron of the cognate hsc70 heat shock Cell, 75, 403-405. premessenger RNA. Cell, 71, 1215-1221. Sumpter,V., Kahrs,A., Fischer,U., Kornstadt,U. and Luhrmann,R. (1992)

4870 Maturation of small nucleolar RNAs

In vitro reconstitution of U1 and U2 snRNPs from isolated proteins and snRNA. Mol. Biol. Rep., 16, 229-240. Terns,M.P. and Dahlberg,J.E. (1994) Retention and 5' cap trimethylation of U3 snRNA in the nucleus. Science, 264, 959-961. Terns,M.P., Lund,E. and Dahlberg,J.E. (1992) 3'-end dependent formation of U6 small nuclear ribonucleoprotein particles in Xenopus laevis oocyte nuclei. Mol. Cell. Biol., 12, 3032-3040. Terns,M.P., Dahlberg,J.E. and Lund,E. (1993a) Multiple Cis-acting Signals for Export of pre-Ul snRNA from the Nucleus. Genes Dev., 7, 1898-1908. Terns,M.P., Lund,E. and Dahlberg,J.E. (1993b) A Pre-export Ul snRNP in Xenopus laevis Oocyte Nuclei. Nucleic Acids Res., 21, 4569-4573. Tollervey,D., Lehtonen,H., Carmo-Fonseca,M. and Hurt,E.C. (1991) The small nucleolar RNP protein NOPI (fibrillarin) is required for pre- rRNA processing in yeast. EMBO J., 10, 573-583. Tollervey,D., Lehtonen,H., Jansen,R., Kern,H. and Hurt,E.C. (1993) Temperature-sensitive mutations demonstrate roles for yeast fibrillarin in pre-rRNA processing, pre-rRNA methylation, and ribosome assembly. Cell, 72, 443-457. Tyc,K. and Steitz,J.A. (1989) U3, U8 and U13 comprise a new class of mammalian snRNPs localized in the cell nucleolus. EMBO J., 8, 3113-3119. Tycowski,K.T., Shu,M.D. and Steitz,J.A. (1993) A small nucleolar RNA is processed from an intron of the human gene encoding ribosomal protein S3. Genes Dev., 7, 1176-1190. Ueland,P.M. (1982) Pharmacological and biochemical aspects of S- adenosylhomocysteine and S-adenosylhomocysteine hydrolase. Pharmacol. Rev., 34, 223-253. Yuan,Y and Reddy,R. (1991) 5' flanking sequences of human MRP/7-2 RNA gene are required and sufficient for the transcription by RNA polymerase III. Biochim. Biophys. Acta, 1089, 33-39. Received on April 18, 1995; revised on May 24, 1995

4871